F. J. Soares INESC Porto/FEUP Smart charging strategies for efficient management of the grid and generation systems Electric Vehicle Integration Into Modern Power Networks 24 September 2010 DTU, Copenhagen
F. J. SoaresINESC Porto/FEUP
Smart charging strategies for efficient
management of the grid and
generation systems
Electric Vehicle Integration Into Modern Power Networks
24 September 2010
DTU, Copenhagen
Summary
1. The Electric Mobility Paradigm
a) Motives for EV adoption
b) Expectable benefits
c) Foreseen problems for electric power systems
d) Predicted EV rollout in some EU countries
2. Conceptual Framework for EV Integration Into Electric Power Systems
a) The EV supplier/aggregator
b) Possible EV charging approaches
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese LV grid
b) Case study B: typical Portuguese MV grid
c) Overall conclusions
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
a) Introduction
b) Case study: Flores Island network (Azores Archipelago)
c) EV motion simulation
d) Monte Carlo Algorithm
e) Results
f) Conclusions
5. Final Remarks
1. The Electric Mobility Paradigm
a) Motives for EV adoption
Extremely volatile oil prices with a rising trend (due to increasing demand)
So
urc
e:
oil-
price.n
et
1. The Electric Mobility Paradigm
a) Motives for EV adoption
High concentration of GHG in the atmosphere (global problem)
So
urc
e:
wik
ipedia
.org
So
urc
e:
wik
ipedia
.org
1. The Electric Mobility Paradigm
a) Motives for EV adoption
High pollution levels in areas with high population density (local problem)
Source: SMH
Source: isiria.wordpress.com
Source: fearsmag.com
1. The Electric Mobility Paradigm
b) Expectable benefits
Reduction of the fossil fuel usage in the transportations sector
Immediate reduction of the local pollution levels
(CO2, CO, HC, NOX, PM)
If EV deployment is properly accompanied by an increase in
the exploitation of renewable endogenous resources
GHG global emissions will be greatly reduced Important
contribution to eradicate the global warming problematic
Source: topnews.in
So
urc
e:
myclim
ate
ch
ang
e.n
et
1. The Electric Mobility Paradigm
b) Expectable benefits
EV capability to inject power into the grid (V2G concept) might be used to
“shape” the power demand, avoiding very high peak loads and energy losses
EV storage capability might be used to avoid wasting “clean” energy
(wind/PV) in systems with a high share of renewables
During the periods when renewable power available
is higher than the consumption
Isolated networks might improve their robustness and safely accommodate a
larger quantity of intermittent renewable energy sources
If EV batteries are efficiently exploited as storage devices
and used to mitigate frequency oscillations
1. The Electric Mobility Paradigm
c) Foreseen problems for electric power systems
Depending on the number of EV present in the grid, the increase in the
power demand will lead to:
• Branches overloading
• Under voltage problems
• Significant increase of the energy losses
• Substation transformers overloading
• Need to invest in new generation facilities to face increasing demand
• Aggravation of the voltage imbalances between phases (for single phase
EV/Grid connections)
1. The Electric Mobility Paradigm
d) Predicted EV rollout in some EU countries
Almost no official information available
Contradictory information from non official sources
Difficult to make accurate network
impact studies
Source: Ricardo plc 2010
Source: Ricardo plc 2010
ACEA - European Automobile Manufacturers' Association
1. The Electric Mobility Paradigm
d) Predicted EV rollout in some EU countries
Types of EV available:
Plug-in Hybrid EV use a small battery
and a generator combined with an ICE
Fuel Cell EV store energy in H2 which
feeds a fuel cell that produces electricity
and heat
Battery EV powered only by electricity,
which requires a large battery pack
2. Conceptual Framework for EV Integration Into Electric Power Systems
a) The EV supplier/aggregator
Single EV do not have enough “size” to participate in electricity markets
If grouped through an aggregator agent, EV might sell several system services
in the markets
The EV suppliers/aggregators:
are completely independent from the DSO
act as an interface between EV and electricity markets
group EV, according to their owners’ willingness, to exploit business
opportunities in the electricity markets
develop their activities along a large geographical area (e.g. a country)
2. Conceptual Framework for EV Integration Into Electric Power Systems
a) The EV supplier/aggregator
EV
supplier/aggregator
structure:Regional Aggregation Unit
Microgrid Aggregation Unit
Microgrid Aggregation Unit
CVC
CVC
CVC
Microgrid Aggregation Unit
MV Level
LV Level
Smart Meter
VC
Smart Meter
VC
Smart Meter
VC
Smart Meter
VC
Smart Meter
VC
Smart Meter
VC
EV Owner
EV Owner
EV Owner
EV Owner
EV Owner
EV Owner
SU
PP
LIE
R/A
GG
RE
GA
TO
R
Regional Aggregation Unit
Microgrid Aggregation Unit
Microgrid Aggregation Unit
CVC
CVC
CVC
Microgrid Aggregation Unit
MV Level
LV Level
Smart Meter
VC
Smart Meter
VC
Smart Meter
VC
Smart Meter
VC
Smart Meter
VC
Smart Meter
VC
EV Owner
EV Owner
EV Owner
EV Owner
EV Owner
EV Owner
• Regional
Aggregation Unit
(RAU) – located at
the HV/MV
substation level and
covering a region
(e.g. a large city) with
~20000 clients
• Microgrid
Aggregation Unit
(MGAU) – located at
the MV/LV substation
level and covering a
LV grid with ~400
clients
PLAYERSCONTROL HIERARCHY
DMS
CAMC
CVC
MGCC
Control
Level 3
VC
RAU
MGAU
TSO
GENCO
DSO
Control
Level 1
Control
Level 2
EV Supplier/AggregatorDis
trib
uti
on
Sy
ste
m
Transmission System
Generation System
Ele
ctr
icity M
ark
et
Op
era
tors
Technical Operation Market Operation
Electric Energy
Electric Energy
Technical Validation of the Market Negotiation (for the transmission system)
Electric Energy
Reserves
Reserves
Parking Parking Battery
Replacement
Battery
Replacement
EV
Owner/Electricity
Consumer
Parking
Facilities
Battery
Suppliers
Electricity
Consummer
Electricity
Consumer
Electric Energy
Controls (in normal system operation) At the level of
Communicates with
Sell offer
Buy offer
Technical validation of the market results
Controls (in abnormal system operation/emergency mode)
Reserves
2. Conceptual Framework for EV Integration Into Electric Power Systems
a) The EV supplier/aggregator
DMS – Distribution Management System CAMC – Central Autonomous Management System MGCC – MicroGrid Central Controller
CVC – Cluster of Vehicles Controller VC – Vehicle Controller
2. Conceptual Framework for EV Integration Into Electric Power Systems
b) Possible EV charging approaches
EV as uncontrollable static loads:
EV owners define when and where EV will charge, how much power they will require
from the grid and the period during which they will be connected to it
EV as controllable dynamic loads:
EV owners give the aggregator the possibility to manage their charging during the
period they are connected to the grid
They only inform the aggregator about the time during which their vehicles will be
connected to the grid and the batteries’ SOC they desire at the end of that same period
EV as controllable dynamic loads and storage devices:
EV are not regarded just as dynamic loads but also as dispersed energy storage
devices
They can be used either to absorb energy and store it or inject electricity to grid,
acting in a V2G perspective
2. Conceptual Framework for EV Integration Into Electric Power Systems
b) Possible EV charging approaches
Charging approaches:
Charging Modes
Uncontrolled
Dumb Charging (DC)
Multiple Prices Tariff (MPT)
Controlled
Smart Charging (SC)
Vehicle-to-Grid (V2G)
2. Conceptual Framework for EV Integration Into Electric Power Systems
b) Possible EV charging approaches
Uncontrolled approaches:
Dumb charging EV owners are completely free to charge their vehicles whenever they want;
electricity price is assumed to be constant along the day
Multiple prices tariff EV owners are completely free to charge their vehicles whenever they
want; electricity price is assumed not to be constant along the day, existing some periods where its
cost is lower
EV
AMM
µG
µG
Storage
Energy absorbed and
charging period of a single EV
EV Charger
Charging starts when
EV is plugged-in
Billing and
tariffsInformation about interruptions
and disconnection orders in
case of grid problems
DSO Aggregator
Power
consumed
MarketResponsible for the
grid technical
operation
2. Conceptual Framework for EV Integration Into Electric Power Systems
b) Possible EV charging approaches
Controllable approaches:
Smart charging active management system where there is an aggregator serving as link
between the electricity market and EV owners; enables congestion prevention and voltage control
V2G mode of operation besides the charging, the aggregator controls the power that EV might
inject into the grid; EV have the capability to provide peak power and to perform frequency control
EV
AMM
µG
µG
Storage
Period during which a single EV will be
connected to the grid and the required
battery SOC at the end of that time
EV Charger
EV is plugged-in and its owner
defines the disconnection hour
and the required battery SOC
Broadcast of information related
with billing, tariffs, set-points to
adjust EV control parameters and
SC/V2G set-points in accordance
with the market negotiations
DSOAggregator
Information about interruptions
and disconnection orders in
case of grid problems
Power
consumed
MarketResponsible for the
grid technical
operation
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
Objectives:
Quantify the maximum percentage of conventional vehicles that can be
replaced by EV, without compromising grid normal operation, using three
different charging approaches:
• Dumb charging
• Dual tariff policy (= multiple prices tariff)
• Smart charging
Compare grid behaviour when subjected to different percentages of EV
and when different charging approaches are implemented
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
Grid architecture:
Semi-urban MV network (15 kV)
Two feeding points voltage 1.05 p.u.
Consumption during a typical weekday
271.1 MWh
Peak load 16.6 MW
0
2
4
6
8
10
12
14
16
18
1 5 9 13 17 21
Co
nsu
mp
tio
n (M
W)
Hour
Total
Household
Commercial
Industrial
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
EV characterization and modelling:
Initially, 635 EV (~5%) were distributed through the grid proportionally to
the residential load installed at each bus
12700 vehicles
Annual mileage 12800 km (35 km/day)
EV assumed charging time 4h
EV fleet considered:
• Large EV 24 kWh 40% of the EV fleet
• Medium EV 12 kWh 40% of the EV fleet
• Plug-in Hybrid EV 6 kWh 20% of the EV fleet
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
Dumb charging and dual tariff policy methodology
Distribute EV through the grid proportionally to the residential power installed in each node
Define the initial share of conventional vehicles replaced by EV
Define, in a hourly basis, the nodal EV load, if no control over charging is imposed (dumb charging mode)
Run a power flow for the current hour
Feasible operating conditions ?
Yes
No
Calculate, in a hourly basis, the total nodal load
End of day was reached ?
No
Yes
Maximum share of EV was reached
Define, in a hourly basis, the nodal conventional load (residential, commercial and industrial) of the grid
Next hour
Increase the share of EV in 1%
Algorithm developed
to quantify the
maximum number of
EV that can be safely
integrated into the
grid with the dumb
charging (without
grid reinforcements)
3. Evaluation of EV Impacts in
Distribution Networks –
Preliminary Studies
a) Case study A: typical Portuguese
MV grid
Smart charging methodology
Algorithm developed to
maximize the number of EV
that can be safely integrated
in the grid with the smart
charging (without grid
reinforcements)
Distribute EV through the grid proportionally to the residential power installed in each node
Define the initial share of conventional vehicles replaced by EV
Define, in a hourly basis, the nodal EV load, if no control over charging is imposed (as in the dumb
charging mode)
Run a power flow for the current hour
Feasible operating conditions ?
Halt the charging
of 2% of the EV
connected in each
node downstream
the problematic
branch
NoYes
Any EV waiting to
resume its charging ?
Record current grid conditions
Calculate, in a hourly basis, the total nodal load
Run a power flow with the new load conditions
Feasible operating conditions ?
Yes
No
Run a power flow with the new load conditions
Feasible operating conditions ?
Resume the charging of the first 5% of EV on
the halted EV list
Yes
Yes
Restore the recorded previous grid conditions
No
End of day was reached ?No
List of EV whose charging
was halted is empty ?
Yes
Maximum share of EV was reached
No
Define, in a hourly basis, the nodal conventional load (residential, commercial and industrial) of the grid
Update the list of EV whose charging was
halted (**)
Update the list of EV whose charging was
halted
Yes
Increase the
share of EV in
1%
Yes
Next hour
Voltage or
congestion problem ?
Halt the charging
of 5% of the EV
connected in the
problematic node
Voltage Congestion
No
Sm
art C
ha
rgin
g
Define the connection period of each EV (*)
(*) The EV connection period was
defined according to the mobility
statistical data gathered for Portugal,
published in [17].
(**) This list is updated and sorted
each cycle, giving priority to EV who
will disconnect first from the grid.
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
Results regarding the maximum allowable EV integration
Dumb charging approach – 10% allowable EV integration
Dual tariff policy – 14% allowable EV integration (considering that 25%
of the EV only charge during the cheaper period – valley hours)
Smart charging strategy – 52% allowable EV integration (considering
that 50% of EV owners adhered to the smart charging system)
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
Scenarios used to evaluate EV impacts in the network 1 power flow for
each hour was performed
Scenario 0 Scenario 1 Scenario 2 Scenario 3 Scenario 4
N.º of Vehicles 12700 12700 12700 12700 12700
EVs % 0% 5% 10% 14% 52%
Hybrid Share - 20% 20% 20% 20%
Medium EV Share - 40% 40% 40% 40%
Large EV Share - 40% 40% 40% 40%
Total Energy consumption (MWh) 277.1 283.2 294.0 301.7 388.1
Dumb
charging
limit
Dual
tariff
limit
Smart
charging
limit
Test
case
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
EV electricity demand with the dumb charging (52% EV penetration):
was calculated taking into account mobility statistical data for Portugal
When people arrive
home from work
0
5000
10000
15000
20000
25000
30000
35000
1 5 9 13 17 21
Po
wer
dem
and
(kW
)
Time (h)
Dumb Charging
EV load
Household load
Total load
0
5000
10000
15000
20000
25000
30000
35000
1 5 9 13 17 21
Po
wer
dem
and
(kW
)
Time (h)
Dumb Charging
EV load
Household load
Total load
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
EV electricity demand with the dual tariff policy (52% EV penetration):
was calculated taking into account mobility statistical data for Portugal
was assumed that 25% of EV owners adhered to this scheme, shifting their EV
charging to lower energy price periods
0
1
2
3
4
5
6
7
8
0
5000
10000
15000
20000
25000
30000
35000
1 5 9 13 17 21
Ele
ctr
icit
y p
rice
Po
wer
dem
and
(kW
)
Time (h)
Dual Tariff Policy
EV load
Household load
Total load
Electricity price
0
1
2
3
4
5
6
7
8
0
5000
10000
15000
20000
25000
30000
35000
1 5 9 13 17 21
Ele
ctr
icit
y p
rice
Po
wer
dem
and
(kW
)
Time (h)
Dual Tariff Policy
EV load
Household load
Total load
Electricity price
When electricity is
cheaper
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
EV electricity demand with the smart charging (52% EV penetration):
was assumed that 50% of EV owners adhered to this scheme, being their
charging controlled by the aggregator
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
1 5 9 13 17 21
Po
wer
dem
and
(kW
)
Time (h)
Smart Charging
EV load
Household load
Total load
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
1 5 9 13 17 21
Po
wer
dem
and
(kW
)
Time (h)
Smart Charging
EV load
Household load
Total load
Avoids peak load
increase
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
Results Changes in load diagrams with 52% of EV penetration
0
5
10
15
20
25
30
35
1 5 9 13 17 21
Lo
ad
(M
W)
Hour
Without EV
Dumb Charging
Dual Tariff Policy
Smart Charging
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
Results Voltages obtained for the worst bus during the peak hour
0,82
0,84
0,86
0,88
0,90
0,92
0,94
0,96
0,98
No Evs 5% Evs 10% Evs 14% Evs 52% Evs
Vo
ltag
e (
p.u
.)
No EVs Dumb charging Dual tariff policy Smart charging
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
Results Worst branch loading obtained during the peak hour
0
20
40
60
80
100
120
140
160
No Evs 5% Evs 10% Evs 14% Evs 52% Evs
Rat
ing
(%)
No EVs
Dumb charging
Dual tariff policy
Smart charging
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
Results Daily losses
0%
1%
2%
3%
4%
5%
6%
7%
0
5
10
15
20
25
30
Without EV 10% EV 14% EV 52% EV
Lo
sse
s re
lative
va
lue
(%
)
Lo
sse
s (M
Wh
)
Losses with no EV (MWh)
Dumb charging losses (MWh)
Dual tariff policy losses (MWh)
Smart charging losses (MWh)
Losses relative value (% of the energy consumption)
0%
1%
2%
3%
4%
5%
6%
7%
0
5
10
15
20
25
30
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
a) Case study A: typical Portuguese MV grid
Results Branches loading overview (peak hour), with 52% EV penetration
No EV Dumb charging
Dual tariff policy Smart charging
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Objectives:
Develop a smart charging strategy to:
1. Maximize the number of EV that can be safely connected into the
grid (without reinforcing it)
2. Minimize the renewable energy wasted (in scenarios where
renewable generation surplus might exist)
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
1st Objective – Maximize the number of
EV that can be safely connected into the
grid (without reinforcing it)
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Grid architecture:
Residential LV network (400 V)
Feeding point voltage 1 p.u.
Feeder capacity 630 kW
250 households
9.2 MWh/day
550 kW peak load
0
20
40
60
80
100
120
1 3 5 7 9 11 13 15 17 19 21 23
% o
f th
e c
on
sum
pti
on
Hour
Total Household Commercial
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
EV characterization and modelling:
Initially, 20 EV (~5%) were distributed through the grid proportionally to
the residential load installed at each bus
375 vehicles
Annual mileage 12800 km (35 km/day)
EV assumed charging time 4h
EV fleet considered:
• Large EV 24 kWh 40% of the EV fleet
• Medium EV 12 kWh 40% of the EV fleet
• Plug-in Hybrid EV 6 kWh 20% of the EV fleet
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Dumb charging and dual tariff policy methodology (same as in case study A)
Distribute EV through the grid proportionally to the residential power installed in each node
Define the initial share of conventional vehicles replaced by EV
Define, in a hourly basis, the nodal EV load, if no control over charging is imposed (dumb charging mode)
Run a power flow for the current hour
Feasible operating conditions ?
Yes
No
Calculate, in a hourly basis, the total nodal load
End of day was reached ?
No
Yes
Maximum share of EV was reached
Define, in a hourly basis, the nodal conventional load (residential, commercial and industrial) of the grid
Next hour
Increase the share of EV in 1%
Algorithm developed
to quantify the
maximum number of
EV that can be safely
integrated into the
grid with the dumb
charging (without
grid reinforcements)
3. Evaluation of EV Impacts in
Distribution Networks –
Preliminary Studies
b) Case study B: typical Portuguese
LV grid
Smart charging methodology
(same as in case study A)
Algorithm developed to
maximize the number of EV
that can be safely integrated
in the grid with the smart
charging (without grid
reinforcements)
Distribute EV through the grid proportionally to the residential power installed in each node
Define the initial share of conventional vehicles replaced by EV
Define, in a hourly basis, the nodal EV load, if no control over charging is imposed (as in the dumb
charging mode)
Run a power flow for the current hour
Feasible operating conditions ?
Halt the charging
of 2% of the EV
connected in each
node downstream
the problematic
branch
NoYes
Any EV waiting to
resume its charging ?
Record current grid conditions
Calculate, in a hourly basis, the total nodal load
Run a power flow with the new load conditions
Feasible operating conditions ?
Yes
No
Run a power flow with the new load conditions
Feasible operating conditions ?
Resume the charging of the first 5% of EV on
the halted EV list
Yes
Yes
Restore the recorded previous grid conditions
No
End of day was reached ?No
List of EV whose charging
was halted is empty ?
Yes
Maximum share of EV was reached
No
Define, in a hourly basis, the nodal conventional load (residential, commercial and industrial) of the grid
Update the list of EV whose charging was
halted (**)
Update the list of EV whose charging was
halted
Yes
Increase the
share of EV in
1%
Yes
Next hour
Voltage or
congestion problem ?
Halt the charging
of 5% of the EV
connected in the
problematic node
Voltage Congestion
No
Sm
art C
ha
rgin
g
Define the connection period of each EV (*)
(*) The EV connection period was
defined according to the mobility
statistical data gathered for Portugal,
published in [17].
(**) This list is updated and sorted
each cycle, giving priority to EV who
will disconnect first from the grid.
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Results regarding the maximum allowable EV integration
Dumb charging approach – 11% allowable EV integration
Smart charging strategy – 61% allowable EV integration (considering
that 50% of EV owners adhered to the smart charging system)
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Scenarios used to evaluate EV impacts in the network 1 three-phase
power flow for each hour was performed
Scenario 0 Scenario 1 Scenario 1
N.º of Vehicles 375 375 375
EVs % 0% 11% 61%
Hybrid Share - 20% 20%
Medium EV Share - 40% 40%
Large EV Share - 40% 40%
Total Energy consumption (MWh) 9.17 9.81 12.74
Dumb
charging
limit
Smart
charging
limit
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Total electricity demand with the dumb and smart charging (61% EV penetration):
The dumb charging curve was calculated taking into account mobility statistical
data for Portugal
The smart charging curve obtained assuming that 50% of EV owners adhered to
this scheme, being their charging controlled by the aggregator
0
200
400
600
800
1000
1 3 5 7 9 11 13 15 17 19 21 23
kW
Hour
Without EVs
Dumb Charging
Smart charging
Feeder capacity
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Results Voltages obtained for the worst bus during the peak hour
0,90
0,91
0,92
0,93
0,94
0,95
0,96
0,97
No EVs 11% - Dumb Charging
11% - Smart Charging
61% - Dumb Charging
61% - Smart Charging
Vo
ltag
e (
p.u
.)
Phase R Phase S Phase T
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Results Worst branch loading obtained during the peak hour
6372
64
124
75
0
20
40
60
80
100
120
140
No EVs 11% - Dumb Charging
11% - Smart Charging
61% - Dumb Charging
61% - Smart Charging
Co
nge
stio
n L
eve
l (%
)
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Results Daily losses
17 11
130
83
0
20
40
60
80
100
120
140
Dumb charging
Smart charging
Dumb charging
Smart charging
Incr
eas
e in
lo
sse
s d
ue
to
EV
s co
nsu
mp
tio
n (
%)
11% EVs 61% EVs
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Results Load imbalance between phases
4,86,0
4,7
14,2 14,0
0
2
4
6
8
10
12
14
16
No EVs 11% - Dumb Charging
11% - Smart Charging
61% - Dumb Charging
61% - Smart Charging
Load
Im
bal
ance
in
th
e M
V/L
V T
ran
sfo
rme
r (%
)
, , , ,
, ,% 100
R S T R S T
MAX MIN
R S T
AVERAGE
P PLI
P
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
2nd Objective – Minimize the renewable
energy wasted (in scenarios where
renewable generation surplus exist)
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Selected scenario A wet and windy day in 2011
Portuguese situation in 2011:
Around 5 GW of wind power + “must run” of the thermal units renewable
energy might be wasted (in low demand periods)
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
1 3 5 7 9 11 13 15 17 19 21 23
P (M
W)
Hour
DER - Hydro Hydro - Run of River Coal
NG Fuel Der - Thermal
Hydro (with reservoir) DER - Wind Demand
Portuguese Generation Profile for a Windy Day in 2011
Installed Capacity (MW)
Hydro - 4957
Thermal - 5820
CHP - 1463
Wind - 5000
Others - 52
Installed Capacity (MW)
Wind energy produced - 51 GWh
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Demand change due to 11% of EV Results obtained for the LV grid were
transposed to the complete electric power system
LV Grid Load Diagram Portuguese Generation Profile
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
1 3 5 7 9 11 13 15 17 19 21 23
P (M
W)
Hour
DER - Hydro Hydro - Run of River
Coal NG
Fuel DER - Thermal
Hydro (with reservoir) DER - Wind
Demand without EVs Demand with EVs - Smart charging
Demand with EVs - Dumb charging
Renewable Energy Wasted!
Wind
Energy
Wasted 31
30
15
0 5 10 15 20 25 30 35
No EVs
Dumb Charging
Smart Charging
%
0
200
400
600
800
1000
1 3 5 7 9 11 13 15 17 19 21 23
kW
Hour
Without EVs
Dumb Charging
Smart charging
Feeder capacity
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Demand change due to 61% of EV Results obtained for the LV grid were
transposed to the complete electric power system
0
200
400
600
800
1000
1 3 5 7 9 11 13 15 17 19 21 23
kW
Hour
Without EVs
Dumb Charging
Smart charging
Feeder capacity
LV Grid Load Diagram National Generation Profile
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
1 3 5 7 9 11 13 15 17 19 21 23
P (M
W)
Hour
DER - Hydro Hydro - Run of River
Coal NG
Fuel DER - Thermal
Hydro (with reservoir) DER - Wind
Demand without EVs Demand with EVs - Dumb charging
Demand with EVs - Smart charging
Large Peak Load Increase!
Wind Energy Wasted
31
26
1
0 5 10 15 20 25 30 35
No EVs
Dumb Charging
Smart Charging
%
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
b) Case study B: typical Portuguese LV grid
Daily CO2 emissions
29 26
11
3031
36
0
10
20
30
40
50
60
70
Without EVs 11% EVs* 61% EVs*
Dai
ly C
O2
em
issi
on
s (k
ton
)
Power system emissions (including: extraction and processing; raw material
transport; and electricity generation)
Light vehicles emissions (well-to-wheel)
*Smart charging
3. Evaluation of EV Impacts in Distribution Networks – Preliminary Studies
c) Overall conclusions
Losses increase as the number of EV rises
Overall GHG emissions decrease as the number of EV rises
Voltages and branches loading worsen as the number of EV increases
~10% is the number of EV that can be integrated with the dumb charging
~15% is the number of EV that can be integrated with the dual tariff policy
When comparing with the dumb charging and with the dual tariff policy, the smart
charging allows:
decreasing grid losses and consequently GHG emissions
improving voltage profiles and branches’ congestion levels
safely integrating 50-60% of EV
avoiding the loss of renewable energy
Results are highly dependent on where and when EV will charge A Monte Carlo
simulation method should be used to obtain more accurate results
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
a) Introduction
The utilization of a Monte Carlo method to perform impact studies is more
adequate allows reducing the uncertainties by running a high number of
different scenarios
This approach allows obtaining average values and confidence intervals for
several system indexes, like buses voltages, branches loading and energy
losses
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
b) Case study: Flores Island network (Azores Archipelago)
Grid architecture:
Isolated MV network
(15 kV)
Typical winter day
consumption 47.55
MWh
2.59 MW peak load
(occurs at 19:30 h)
Average power factor
0.77
Island light vehicles
fleet 2285 vehicles
2 scenarios studied
25% and 50% EV
penetration
1
2 7 8 17 41
3 9 42
4 19 43
5 11 20 29 31 44
6 12 21 32 37
13 22 33 38
14 23 34 39
15 24 40
16 25
26
27
Swing Bus
Thermal Power Plant Hydro Power Plant
Wind Farm
28
18
35
4530
24 Bus
Load
Power Plant
Line
10
36
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
c) EV motion simulation
EV movement along one day was simulated using a discrete-time non-Markovian
process to define the states of all the EV at each 30 minutes interval (48 time instants)
In each time instant, EV can be in four different states: in movement, parked in
industrial area, parked in commercial area, parked in residential area
The EV state for each time instant is defined according to the probabilities specified for
that time instants and according to the discrete-time non-Markovian process
In Movement
Parked in
Industrial Area
Parked in
Residential Area
Parked in
Commercial Area
𝒕 = 𝒏
𝑃𝐶→𝑀𝑡=𝑛
𝑃𝐶𝑡=𝑛
In Movement
Parked in
Industrial Area
Parked in
Residential Area
Parked in
Commercial Area
𝒕 = 𝒊
𝑃𝐶→𝑀𝑡=𝑖
𝑃𝐶𝑡=𝑖
In Movement
Parked in
Industrial Area
Parked in
Residential Area
Parked in
Commercial Area
𝑃𝑀𝑡=1
𝑃𝑀→𝑅𝑡=1
𝑃𝑅→𝑀𝑡=1
𝑃𝑀→𝐼𝑡=1 𝑃𝐼→𝑀
𝑡=1
𝑃𝑀→𝐶𝑡=1
𝑃𝐶→𝑀𝑡=1
𝑃𝑅𝑡=1 𝑃𝐼
𝑡=1 𝑃𝐶𝑡=1
𝒕 = 𝟏
𝑃𝐼𝑡=𝑛 𝑃𝑅
𝑡=𝑛
𝑃𝐼𝑡=𝑖 𝑃𝑅
𝑡=𝑖
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
c) EV motion simulation
The state transition probabilities applied were determined by analyzing the common traffic
patterns of Portuguese drivers
It was gathered information about the number of car journeys made per each 30 minutes
interval, along a typical weekday, as well as the journey purpose and its average duration
With this data, it was possible to define the probabilities of an EV reside in a given state at a
given time instant
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
c) EV motion simulation
Define EV location for parked EV:
all bus loads were classified as industrial, commercial or residential
the probability of an EV be located at a specific bus was calculated with the
following equations:
𝑃𝐵𝑢𝑠 𝑘𝑅 =
𝐿𝑜𝑎𝑑𝐵𝑢𝑠 𝑘𝑅
𝐿𝑜𝑎𝑑𝑅 𝑃𝐵𝑢𝑠 𝑘𝐼 =
𝐿𝑜𝑎𝑑𝐵𝑢𝑠 𝑘𝐼
𝐿𝑜𝑎𝑑𝐼 𝑃𝐵𝑢𝑠 𝑘𝐶 =
𝐿𝑜𝑎𝑑𝐵𝑢𝑠 𝑘𝐶
𝐿𝑜𝑎𝑑𝐶
4. Evaluation of EV Impacts in
Distribution Networks – A Monte
Carlo Method
d) Monte Carlo algorithm
1. Make the initial characterization of all the EV:
• initial state
• the bus they are initially located
• battery capacity (kWh)
• slow charging rated power (kW)
• initial SOC (%)
• energy consumption (kWh/km)
• owners’ behaviour
Ind
exe
s
up
da
te
Sa
mp
le g
en
era
tion
an
d e
va
lua
tion
Define EV initial conditions (initial state, bus, battery capacity, slow charging rated
power, initial SOC, energy consumption and driver behaviour)
Draw EV states and the buses where “parked” EV are located, for the next time
instant
Determine the new load at each bus
Power flow analysis
End of the day was reached ?
Monte Carlo finishing criteria was met ?
No
Compile results: power demand, voltages, branches loading, energy losses, peak
power, number of voltage and branches ratings violations
Yes
Yes
EV battery SOC < 30% ?
Yes
Update EV batteries SOC
Update of grid technical indexes and vehicle usage indicators in a hourly and daily
basis
Yes
No
What is the EV driver behaviour ?
EV is parked in
residential area ?
Yes
EV is parked in
residential area ?
EV arrived home from the
last journey of the day ?
Yes
EV starts charging
EV do not charge
No
No No
EV charge
only when
it needs EV charge
whenever
possible
EV charge at the
end of the day or
whenever is
convenient and the
driver has time
No
GAUSSIAN DISTRIBUTIONS FOR INITIAL EV CHARACTERIZATION
Average Standard
deviation
Maximum
value
allowed
Minimum
value allowed
Battery capacity (kWh) 24.73 17.19 85.00 5.00
Slow charging rated power
(kW) 3.54 1.48 10.00 2.00
Energy consumption
(kWh/km) 0.18 0.12 0.85 0.09
Initial battery SOC (%) 50.00 25.00 85.00 15.00
DRIVERS’ BEHAVIOURS CONSIDERED
Percentage of the
responses
EV charge at the end of the day 33%
EV charge only when it needs 23%
EV charge whenever possible 20%
EV charge whenever is convenient and the driver has time 24%
30% SOC
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
d) Monte Carlo algorithm
2. Samples generation:
• Simulate EV movement along one typical weekday define EV states
• Attribute a bus location to parked EV
• Update battery SOC for EV in movement:
o if an EV was in movement in time instant t and its battery SOC went below a
predefined threshold (assumed to be 15%) in time instant t+1, it was considered that
the EV would make a short detour to a fast charging station for recharging purposes
o the fast charging was assumed to be made during 15 minutes with a power of 40 kW
o the fast charging station was considered to be installed in bus 12, as this is located
near one of the more populated areas of the island, with a high number of potential
clients
• Compute the total amount of power required from the network, discriminated per bus and
per time instant
GAUSSIAN DISTRIBUTIONS FOR EV MOVEMENT CHARACTERIZATION
Average Standard
deviation
Maximum
value
allowed
Minimum
value allowed
Travelled distance in
common journeys (km) 9.01 4.51 27.03 0.90
Travelled distance to fast
charging station (km) 4.51 2.25 13.52 0.45
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
d) Monte Carlo algorithm
3. Samples evaluation:
• Made by running a power flow for each time instant and by gathering information about:
o Voltage profiles
o Power flows in the lines
o Energy losses
o Highest peak load
4. Terminating the Monte Carlo process 2 criteria used:
• Number of iterations 10000
• Variation in the last 10 iterations of the aggregated network load variances (of each one of
the 48 time instants) < 1𝑒−4
∆𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒 = 𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒𝑗𝑖 − 𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒𝑗−10
𝑖 < 1𝑒−4
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
e) Results
Power demand:
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
e) Results
Voltage profile of one feeder (buses 17 to 27):
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
e) Results
Network voltage profiles for the highest peak load identified:
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
e) Results
Voltage lower limit violation probability:
𝑃𝑉.𝑙𝑜𝑤𝑒𝑟 𝑙𝑖𝑚𝑖𝑡 𝑣𝑖𝑜𝑙𝑎𝑡𝑖𝑜𝑛𝐵𝑢𝑠 𝑘 =
𝑉. 𝑙𝑜𝑤𝑒𝑟 𝑙𝑖𝑚𝑖𝑡 𝑣𝑖𝑜𝑙𝑎𝑡𝑖𝑜𝑛𝑠𝐵𝑢𝑠 𝑘
𝑁𝑟. 𝑖𝑡𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 × 48× 100
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
e) Results
Branches loading:
No EV
25% EV
50% EV
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
e) Results
Average daily energy losses:
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
e) Results
Evolution of the network load variances with the highest variation rate:
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
e) Results
Network load variances of the 48 time instants:
4. Evaluation of EV Impacts in Distribution Networks – A Monte Carlo Method
f) Conclusions
The simulation platform developed proved to be very efficient in performing a realistic
evaluation of the impacts that result from a massive integration of EV in distribution networks
Allows:
evaluating the steady state operating conditions of the grid
identifying the most critical operation scenarios and the network components that are
subjected to more demanding conditions and that might need to be upgraded
The island network is very robust is capable of integrating a large number of EV without
the occurrence of lines overloading and voltage limits violations (~25%)
With 50% of EV a large number of voltage violations were registered efficient
mechanisms to manage EV charging (smart charging) are required to avoid making large
investments in network reinforcements
For large EV integration scenarios, losses will become a very important issue for system
operator their value grows:
58% from the scenario without EV to the one with 25% of EV
140% from the scenario without EV to the one with 50% of EV
Energy losses might be greatly reduced by using an EV smart charging strategy
5. Final Remarks
EV integration in interconnected systems:
Due to the reduced energy consumption and capability of providing
services to the grid, it is impossible to EV participate in the markets
individually EV suppliers/aggregators must exist for this purpose
Even under the EV supplier/aggregator management, EV might still create
several problems in distribution networks A grid monitoring mechanism
must exist (independent from the aggregator and headed by the DSO), with
the capability of manage EV charging, in order to avoid those problems
EV integration in small isolated systems:
As usually these systems do not have an electricity market, EV
suppliers/aggregators are not needed Only the grid monitoring mechanism
controlled by the DSO must exist
5. Final Remarks
EV integration limitations:
without any control actions over EV charging (dumb charging), it is
impossible to integrate a large number of EV in common electricity networks
network reinforcements are required
if EV charging is controlled (smart charging), even in accordance with their
owners requirements, a larger number of EV might be integrated without
investments in grid reinforcements
nonetheless, if the number of EV keeps growing, there will be a moment in
time where reinforcement will be inevitable, even when the smart charging is
applied…
5. Final Remarks
Network impacts As the number of EV rises:
losses increase
overall GHG emissions decrease
voltages and branches loading worsen
Smart charging vs. Dumb charging:
decrease grid losses and, consequently, GHG emissions
improve voltage profiles and branches loading
allow the integration of a higher number of EV without reinforcements
allow an effective exploitation of renewable generation surplus (when such
problem exists)